Barbara B. Sherwin, Ph.D., and Morrie M. Gelfand, M.D.
From the Department of Psychology, McGiU University, and the Department of Obstetrics and Gynecology, The Sir Mortimer B. Davis— Jewish General Hospital, McGill University.
Supported m part by., a grant from the Planning and Priorities Committee, The Sir Mortimer B, Davis—Jewish General Hospital, Montreal, Quebec, Canada.
Presented at the Fortieth Annual Meeting of The Society of Obstetricians and Gynaecologists of Canada, Montreal, Quebec, Canada,. June 19-23, 1984.
Reprint requests: Dr. B. Sherwin, McGill University, Department of Psychology, 1205 Docteur Penfield'Ave., Montreal, Quebec, Canada H3A 1B1.
The investigation of estrogen and/or androgen administration on physical and psychological symptoms In the surgical menopause was carried out in a prospective, double-blind, crossover design. When patients who received either a combined estrogen-androgen drug or androgen alone were compared with those who received estrogen alone or placebo, energy level, well-being, and appetite were increased (p < 0.01). The androgen-containing preparations also induced lower somatic, psychological, and total scores on the menopausal index. Superior functioning in the androgen-treated groups occurred in association with higher plasma testosterone levels during the treatment phases (p < 0.01). These data suggest that reduced levels of circulating testosterone subsequent to bilateral oophorectomy may play an important role in the development of physical and psychological symptoms that are frequent sequelae of this surgical procedure. (AM J OBSTET GYNECOL 1985;151:153-60.)
Key words: Surgical menopause, estrogen and androgen, androgen parenteral therapy.
The precipitate decline in plasma estrogen levels following bilateral oophorectomy in premenopausal women has been well documented,1,2 Less frequently acknowledged is the significant decrease in plasma testosterone levels subsequent to removal of both ovaries in this same population.3 The observation that testosterone levels in ovarian vein blood are greater in early postmenopausal women than in reproductive-aged women4 implies that patients who need to undergo bilateral oophorectomy are subsequently deprived of an important source of androgen production. Despite these known endocrine changes and the significant incidence of symptoms following total abdominal hysterectomy and bilateral salpingo-oophorectomy,5 little evidence is available with regard to the functional role of androgen in the woman.
In order to evaluate possible differential effects of estrogen and androgen ori symptoms in surgically menopausal women, a prospective, double-blind, crossover study was undertaken. The objectives of this investigation were twofold: (1) to independently assess the clinical response of somatic and psychological symptoms to exogenous estrogen and androgen administered singly as well as in combination and (2) to address the theoretical issues regarding physical and behavioral effects of the sex steroids in women.
Materials and methods
Subjects
The participants in this study were premenopausal women who required a total abdominal hysterectomy and bilateral salpingo-oophorectomy for reasons other than malignant disease. They were required to be in a stable heterosexual relationship for at least 2 years and to have had no past history of psychological disturbances for which treatment had been obtained. A further requirement was that they be in a state of good general health with no known contraindications to hormone replacement therapy. Informed, signed consent was obtained from 43 women who met these criteria by means of a form approved by both a hospital and a university ethics committee. Ten women who needed to undergo total abdominal hysterectomy but in whom the ovaries were to be retained were recruited to serve as an additional control group, Between September, 1980. and February, 1983, each patient who was scheduled for total abdominal hysterectomy and bilateral salpingo-oophorectomy from the practice of the chief of the department of obstetrics and gynecology of a large university teaching hospital was leferred to the study. Subjects were told that they would be participating in “an investigation of estrogen and androgen on physical and psychological functioning.” They were further informed that there was a one-in-four chance that they would receive a preparation “with no active hormone” and that their medication might be changed during the course of the study. However, subjects had no way of knowing when that would occur or. indeed, if it would ever occur.
Procedure
The study was a five tgroups)-by—lour (time periods) repeated measures design. Before operation, patients were randomly assigned to either a combined estrogen-androgen group, an estrogen-alone group, an androgen-alone group, or a placebo group. The control hysterectomy group described previously constituted the fifth group. Thus there were two control groups in the design—one to control for the effects of an altered endocrine status consequent to bilateral salpingo-oophorectomy (placebo group) and the second to control for changes that may have been due to the surgical procedure itself (control hysterectomy group).
Following 1 month of baseline monitoring, patients underwent operation. Intramuscular injection of a hormonal preparation or placebo was administered monthly for the first 3 months (treatment 1). At the beginning of the fourth postoperative month, all patients received an injection of placebo. Women were then randomly crossed over to one of the four treatments they had not experienced during treatment 1 for an additional 3 months (treatment 2). Therefore, the entire experimental procedure encompassed a total of 8 months per patient. Fig. 1 contains a diagramatic representation of the experimental design.
Figure 1. Experimental design.
Drug dosages
In order to ensure that patients in the hormone treatment groups would receive equal amounts of estrogen and androgen per dose, the molecular weights of the salts to which each of the estrogen and androgen preparations was bound were calculated. This sum was then subtracted from the weight of the total compound. With 1 ml of the combined estrogen- androgen drug used as the standard, equivalent doses of the sex steroids contained in the oilier preparations were computed. Accordingly, it was determined that 0.63 ml of the estrogen-alone drug and 0.48 ml ot the androgen-alone drug contained amounts of estradiol and testosterone that were equivalent to amounts of these hormones in 1 ml of the combined preparation. While it is recognized that different esters have different durations of action, there did not appear to be any way to equate these except to calculate free estradiol. Moreover, it has been shown that other sex steroid depot preparations induce maximal plasma hormone levels 1 to 4 days following injection and that the half-lite is 14 days.6 The estrogen-androgen combined drug (Climacteron), 1 ml, contains testosterone enantiiate benzilic acid hydrozone, 150.0 mg; estradiol dieuan- thate, 7.5 mg; and estradiol benzoate, 1.0 mg. The estrogen drug (Delestro'gen), I ml, contains estradiol valerate, 10.0 mg. The androgen drug (Delatcstryl), 1 mV contains testosterone enanthate, 200.0 mg. The placebo consisted of sesame oil, 0.5 ml. After operation, drugs were given intramuscularly at 28-day intervals.
Plasma hormone levels
Venous blood samples for measurement of plasma levels of total estrogens (estradiol and estrone) and testosterone were obtained between 9 and 11 AM at four different times during the study. The first sample was taken approximately 1 month before operation; the second and fourth, 72 hours following the third injection in both treatment phases; and the third, on the last day of the placebo month that fell in between. The blood was immediately centrifuged and the plasma stored at —20° C. Total plasma estrogens were measured by radioimmunoassay with the use of Endocrine Sciences antiserum No. E- 17-94. The extraction of protein before the addition of antiserum ensured that free hormone was being assayed. The assay was sensitive to the level of 12.5 pg/ ml. Plasma testosterone levels were measured with the Covalent-Coat Radioimmunoassay Kit. The Testosterone (125 I) Assay detects the total unconjugated form of this steroid. All samples from each patient were measured twice in the same assay at the conclusion of the study. The control hysterectomy group underwent the identical experimental procedure except for the administration of monthly injections.
Material
Daily menopausal rating scale
Somatic and psychological symptoms were monitored daily by means of the daily menopausal rating scale. This instrument, devised and tested in a previous study,7 was found to be a reliable and sensitive index of drug effects. Each symp-tom or behavior appears on an interval rating scale that has a range of 0 to 7. The poles of each scale have a verbal description of the symptom or behavior in question. Measures of energy level, well-being, appetite, and sleep quality reported here were monitored and quantified by means of the scale. Patients were given 30 questionnaires and 30 stamped, self-addressed envelopes at the beginning of each month and were instructed to fill in and mail a questionnaire daily. The 8-month duration of the study, therefore, yielded data on 240 days per patient arid 12,720 patient days in all.
Menopausal index
The menopausal index, revised by Netigarten and Kraines,8 was originally published by Blau et al.9 It lists 26 symptoms most often reported by clinicians and by women themselves as being typical or frequent complaints at menopause. Symptoms are categorized into three constellations—somatic (nine symptoms), psychosomatic (five symptoms), and psy-chological (12 symptoms) as seen in Table I. Because the scoring system that assigns differential weights to the various symptoms is not empirically based,9 in this study, patients were required to rate the frequency and severity of each symptom on a bipolar rating scale that had a range of 0 to 7. The low end of the scale for each symptom was described as “almost never” while the extreme end stated “very often,” Patients filled out the menopausal index at four points during the study immediately prior to venous blood collection.
Table I. Menopausal index.
Statistical analyses
The hormonal and behavioral data were analyzed with the Computer Programs analysis of variance for groups x time blocks. For each item on the daily menopausal rating scale, the data analyzed were the mean of the preoperative baseline, the mean of the first 2 weeks of each treatment month, and the mean of the last 2 weeks of the intervening placebo month. In addition, analysis of covariance was also carried out in order to eliminate possible effects of initial group differences in baseline scores. Post hoc pairwise comparisons were performed with use of the Tukey test. Statistical significance reported here is based on results of the latter analyses.
Results
Subject characteristics
There were 12 patients in the combined estrogen-androgen group, 11 in the estrogen-alone group, 10 in the androgen-alone group, 10 in the placebo group, and 10 in the control hysterectomy group. Pathology reports confirmed that two patients had endometriosis; one was in the estrogen- alone group and the other in the placebo group. The rest of the sample (51 women) underwent operation for benign uterine fibroids. Five subjects who met the selection criteria and who were initially interviewed did not continue in the study. Two women decided to defer operation, one refused to comply with the experimental procedure alter only 1 week, and a fourth was eliminated after the first postoperative month because she withdrew her acceptance of the random assignment procedure. The fifth patient was found to have an unsuspected carcinoma of the ovary at the time of operation and was thus eliminated from the study. All other subjects remained in the study through its entirety and each provided a complete set of observations.
Subject characteristics measured at baseline showed that patients in tile control hysterectomy group were significantly younger than those in the four treatment groups (50.3 ± 2,2 versus 45.8 ± 3.3 years). This was an expected finding in view of the fact that age is the major criterion that guides the decision to remove or retain normal-appearing ovaries at. the time of total abdominal hysterectomy undertaken for benign con-ditions. Subjects in ail five groups were homogeneous with respect to occupational status and years of school. I here was, however, a wide range in these variables within groups which increased the generalizability of the findings. Ail subjects had normal personality profiles10 and satisfactory marital relationships.11
Total plasma estrogen and testosterone levels
Mean plasma levels of total estrogen (estrone and estradiol) and testosterone at the four test times appear in Table II. 1 he imerassay coefficient of variation of total es- trogen was 9%. At baseline the younger control hysterectomy group had significantly higher total estrogen levels than the four treatment groups (p < 0.01), whose levels did not differ from each other. This was the only measure on which there was any significant difference between the five groups at baseline, which testifies to the success of the randomization procedure. At 72 hours following injection in both treatment phases, all groups had higher total estrogen levels than the placebo group (p < 0.01). Moreover, total estrogen levels of the estrogen-alone group were significantly higher during treatment than those of the androgen-alone and the control hysterectomy groups (p < 0.01) but were not different from those of the estrogen-androgen group. At the end of the placebo month, the control hysterectomy group with intact ovaries had significantly higher total estrogen levels than the four groups of women who had undergone oophorectomy. In these latter women the levels fell to the lower limit of assay detectability, which indicates that the estrogen administered during treatment 1 had been completely metabolized. In all cases, total estrogen levels of subjects who received the same drug did not differ when treatment I and treatment 2 values were compared.
Table II. Mean plasma levels of total estrogen (estrone and estradiol) and testosterone at the four test times.
The interassay coefficient of variation of plasma testosterone levels was 10%. During both treatment phases, the estrogen-androgen group had significantly higher testosterone levels than the estrogen-alone, placebo, and control hysterectomy groups (p < 0.01), but values did not differ significantly from those of the androgen-alone group. By 8 weeks after injection, testosterone values in steroid-treated groups also fell to levels below those of the control hysterectomy group arid levels of the androgen-alone, estrogen-alones and placebo groups were significantly lower, than both their own baseline values and control hysterectomy group, values (p < 0.05). Ortce again, testosterone levels of subjects who received the same drug did not ,differ when group means of the two treatment phasesi.weR compared. Mot covet In imon-gioup dilfciences in both hormone levels at tie timcnt 2 wete i te same n those found during treatment 1.
Energy level and well-being
Hormonal effects on these two measures, thought to reflect how patients were feeling in general, were very similar. In each of the three treatment months of both phases, women in the estrogen-alone and placebo groups reported significantly lower ratings of energy level and well-being than did the control hysterectomy group, with intact ovaries, and those who received either of the androgen- containing preparations (p < 0.01). Moreover, in treatment months 1, 3, 5, 6, and 7, estrogen-alone and placebo group scores were lower than those of the other three groups (p < 0.01). When hormones were withdrawn during the placebo month, the control hysterectomy group with intact ovaries attained higher scores on both measures than did all women who had undergone oophorectomy (p < 0.0.1) coincident with the higher plasma hormone levels at that time, It can be seen in Figs. 2 and 3 that there was a decrease, in energy level and well-being from baseline to the first postoperative month in four groups (p < 0.01); only the patients who received androgen alone maintained energy level and well-being ratings in the immediate postoperative period. Furthermore, by the end of the seventh postoperative month, energy level and well-being scores of the control hysterectomy group and both groups that received androgen (estrogen-androgen and androgen alone) exceeded the preoperative baseline scores (p < 0.01) whereas scores of the estrogen- alone and placebo groups did not differ significantly from baseline values at that time.
Figure 2. Energy level for 24 hours. The treatment groups are: Estrogen-androgen group (E-A, Δ), estrogen-alone group (E, ◊), androgen-alone group (A, □), placebo group (PL, o), anti contiol hysterectomy group (CON, •).
Figure 3. Well-being per 24 hours. The treatment groups are as in Fig. 2.
Appetite
Analysis of daily ratings of appetite quality shown in Fig. 4 found that in each of the treatment months, scores of the estrogen-alone, placebo, and control hysterectomy groups were significantly lower than those of the androgen-alone group (p < 0.01) and lower than those of the estrogen-androgen group as well in months 1, 2, and 7 (p < 0.01). The lower estrogen-alone, placebo, and control hysterectomy group appetite ratings occurred in association with their lower testosterone levels in both treatment phases. During the placebo month: scores of the androgen-alone group decreased significantly compared to the treatment values (p < 0.01). By the third month in both treatment phases, appetite ratings were higher than preoperative baseline scores in only the estrogen-androgen and the androgen-alone groups (p < 0.01).
Figure 4. Appetite quality. Treatment groups as in Fig. 2.
Sleep quality
Analysis of daily ratings of sleep quality showed that there were no significant differences in any of the groups throughout the course of the investigation.
Menopausal index
Somatic symptoms
Lower scores on subtests of the menopausal index indicate a lower frequency and severity of symptoms. Fig. 5 shows that during both treatment phases somatic symptom scores of the estrogen- androgen, androgen-alone, and control hysterectomy groups were lower than those of the estrogen-alone and the placebo groups (p < 0.01), whereas during the placebo month all women who had undergone oophorectomy reported higher scores than the control hysterectomy group (p < 0.01). Furthermore, within-group analyses showed that all steroid-treated groups expe- rtenced a significant increase in scores during the placebo month compared to the treatment phase scores; However, scores of patients in the placebo gtoup increased significantly and consistently after operation compared to the preoperative baseline values (p < 0.01) whereas those of the control hysterectomy group remained stable at low levels throughout the course of the study.
Figure 5. Mean somatic symptom scores. Treatment groups as in Fig. 2.
Psychosomatic symptoms
There were no significant changes in psychosomatic symptom scores in any of the groups across time.
Psychological symptoms
As seen in Fig. 6, psychological symptom scores of the estrogen-alone and placebo groups were significantly higher than those of the estrogen-androgen, androgen-alone, and control hysterectomy groups during both treatment phases (p < 0.01). Moreover, during the intervening placebo month, scores of the control hysterectomy group were lower than those reported by the estrogen-androgen group (p < 0.05) and by the estrogen-alone and placebo groups (p < 0.01). Analysis of within-group changes across time found that baseline psychological symptom scores were lower than treatment phase scores in the estrogen-alone and placebo groups jp < 0.01) whereas both androgen groups and the control hysterectomy group attained lower scores during the treatment phases compared to the baseline values ip < 0.01).
Figure 6. Mean psychological symptom scores. Treatment groups as in Fig. 2.
Total scores
Composite scores of all items on the menopausal index appear in Fig. 7, The estrogen-androgen, androgen-alone, and control hysterectomy groups attained lower total scores during both treatment phases than the estrogen-aione and placebo groups (p < 0.01). However, during the placebo month, total scores of the control hysterectomy group were lower than those of all women who had undergone bilateral salpingo-oophorectomy (p < 0.01). Furthermore, in contrast to the lower scores attained by the estrogen-androgen and androgen-alone groups dining treatment compared to the baseline and placebo phase scores (p < 0.01), the estrogen-alone and placebo groups reported higher total scores on this measure at all postoperative testing times compared to the preoperative values (p < 0.01).
Figure 7. Mean total scores. Treatment groups as in Fig. 2.
Comment
Results of this study unequivocally demonstrated dif-ferential responses of physical and psychological symptoms to estrogen and/or androgen administration in surgically menopausal women. In view of the fact that patients in this investigation were generally healthy and functioning well preoperatively, these data also provide interesting information with regard to the incidence of symptoms in treated and untreated surgically menopausal patients.
The postoperative decrease in both total estrogen and testosterone levels in women who have undergone oophorectomy reflected in the placebo phase hormone assays is consistent with data reported previously.1,3 The single exception was the absence of a statistically significant decline in the testosterone levels of the estrogen-androgen group by the end of the placebo month. It may be that endogenous testosterone levels did not decrease in this group subsequent to operation. However, an alternative hypothesis is that the half-life of the combined preparation is longer than the half-lives of the other drugs, and thus the dose administered had not been completely metabolized by the eighth week following the last injection. Studies on the pharmacodynamics of the combined preparation currently underway will provide a basis for resolving this issue.
Coincident with the observed increases in circulating testosterone levels in the estrogen-androgen and an-drogen-alone groups during treatment was enhancement of energy level, well-being, and appetite reported by these patients. The androgenic effect on these behaviors was likely a reflection of the anabolic12 and energizing properties13 of this steroid. The observation that androgen alone had a more profound effect on energy level, well-being, and appetite relative to the combined drug at several points during treatment can be explained by changes in sex hormone—binding glob-ulin concentrations induced by the sex steroids. It is known that both estrogen and androgen bind to sex hormone-binding globulin in plasma and that only the tree unbound portion has potential for biologic activity14. Because estrogen increases sex hormone—binding globulin levels and androgen reduces them,15 it is possible to infer that the administration of androgen alone in the presence of low levels of endogenous estrogen resulted in a higher percentage of circulating free testosterone in the androgen-alone group compared to the estrogen-androgen group. The relative increase in free testosterone in the androgen-alone group may thus have accentuated the androgenic effects on be havior observed. This hypothesis merits testing by mea-surement of sex hormone-binding globulin subsequent to the administration of both of these drugs.
That exogenous estrogen alone was not more effective than placebo on energy level, well-being, and appetite was clearly demonstrated. Moreover, the finding that scores of the younger control hysterectomy group with intact ovaries were higher than their own baseline values by the end of the seventh postoperative month suggests that preoperative scores may have been depressed in all groups because of psychological factors related to impending operation. If this were the case, the stability in estrogen-alone and placebo group scores following operation may be indicative of an actual postoperative deterioration in their functioning. Studies that have reported increased well-being in postmenopausal women as a function of estrogen administration have been single-blind16 or have used subjects undergoing a natural menopause17 so that endogenous testosterone levels were likely to be unchanged. Other authors,18,19 however, have similarly failed to find a difference between effects of estrogen and placebo on well-being in the postmenopause. It is important to note that in this study increased energy level and well-being did not occur secondary to relief of somatic symptoms since it was found that androgen alone and placebo were equally ineffective in alleviating hot flushes in these same women.20 Moreover, no side effects of exogenous androgen were either observed or reported. Specifically, there were no signs of virilization in the two androgen groups during either of the 3-month treatment phases.
The superior efficacy of the androgen-containing preparations on somatic, psychological, and total scores of the menopausal index may also be related to the anabolic and energizing properties of this sex steroid. Furthermore, the presence of androgen receptors in the hypothalamus, pituitary, and limbic system21 allows for the possibility that this hormone may have been exerting its effects on behavior centrally. Additional evidence for a causal relation between these symptoms and circulating levels of sex steroids derives first from the observation that, when hormones were withdrawn during the placebo month, scores of all oophorectom- ized subjects on the three subscales of the menopausal index increased compared to the treatment phase scores. Second, scores of the placebo group increased consistently and significantly during the postoperative course compared to preoperative values in association with a decrease in circulating total estrogen and testosterone levels. In contrast, scores of the intact control hysterectomy group on the menopausal index maintained stability throughout the study concomitant with stability in sex steroid levels across time.
The consistent clinical superiority of the androgen - containing drugs in this study strongly implies that reduced levels of circulating testosterone may indeed play a critical functional role in the development of a myriad of disturbances that often arise in premenopausal women subsequent to bilateral salpingo-oophorectomy. The clinical implication of these findings is that administration of both sex steroids to this population may militate against symptoms that are a consequence of the sudden and drastically altered endocrine status induced alter operation. These results need to be confirmed both by replication of this study and by investigation of the long-term effects of combined hormone replacement therapy in surgically menopausal women.
These data serve as well to point out a methodologic issue in menopause research. It was noted that the effect of estrogen on hot flushes, the cardinal menopausal symptom, was obscured when it was grouped together with other symptoms, as is the case in the menopausal index. This finding underlines the importance of investigating menopausal symptoms independently until we derive an empirical basis for clustering them.
In summary, results of this study showed that pre-menopausal women who received androgen alone or a combined estrogen-androgen drug following total abdominal hysterectomy and bilateral salpingo-oophorectomy reported fewer somatic and psychological symptoms than those who received estrogen alone or placebo. Furthermore, the androgen-containing preparations induced levels of functioning that were similar to those of younger women with intact ovaries. Bearing in mind that neither androgen alone nor placebo alleviated hot flushes,20 the necessity for administering estrogen as well seems clear. These data provide strong support for the conclusion that a combined estrogen- androgen regimen may serve to enhance the quality of life of women who have experienced a surgical menopause.
Finally, it was observed in this study that menopausal symptoms are dissociable, under different hormonal control, and some, such as sleep quality and psychosomatic symptoms, independent of circulating levels of the sex steroids. In order to further investigate the specificity of hormone-behavior relationships, future research in this area ought to focus on the response of individual symptoms to the administration of each of the sex steroids as well as to their combined effect
References
Vermeulen A. Sex hormone status of the postmenopausal woman. Maturitas 1980;2:81.
Chakravarti S, Collins WP, Newton JR, Studd JWW. Endocrine changes and symptomatology after oophorectomy in premenopausal women. Br j Obstet Gynaecol 1977:84:769.
Vermeulen A. The hormonal activity of the postmenopausal ovary. J Clin Endocrinol Metab 1976;42:247.
Botella-LlusiàJ, Oriol Boosch A, Sanchez Garrido F, Tres- guerres JAF. Testosterone and 17|3-oestradiol secretion of the human ovary. II. Normal postmenopausal women, postmenopausal women with endometrial hyperplasia, and postmenopausal women with adenocarcinoma of the endometrium. Maturitas 1980;2:7.
Dennerstein L, Wood C, Hudson B, Burrows G. Clinical features and plasma hormone levels after surgical menopause. Aust NZ J Obstet Gynecol 1978; 18:202.
Dusterberg B, Wendt H. Plasma levels of déhydroépian-drostérone and 17p-estradiol after intramuscularadmin- istraiion of Gynodian-Depot in 3 women. Horm Res 1983; 17:84.
Sherwin BB, Brender W, Gelfand MM. Mood and behavior changes in menopausal women receiving gonadal hormones or placebo. In: Stark-Adamec C, ed. Sex roles: origins, influences and implications for women. St. Albans, Vermont: Eden Press, 1980:190.
Neugarten B, Kraines RJ. Menopausal symptoms in women of various ages. Psychosom Med 1965;27:266.
Blatt MGF, Wiesbader Y, Kupperman HS. Vitamin E and climacteric syndrome. Arch Intern Med 1953;91:792.
Eysenck HJ, Eysenck SBG. Eysenck personality inventory manual. San Diego: Educational and Industrial Testing Service, 1958.
Locke HJ, Wallace KM. Short marital adjustmentand prediction tests. Marriage Fam Liv 1959;21:251.
Johnson LC, O’Shea JP. Anabolic steroids: effects on strength development. Science 1969; 164:957.
Rose RM. The psychological effect of androgens and estrogens. A review. In: Shader RI, ed. Psychiatric complications of medical drugs. New York: Raven Press, 1972: 251.
Mercier-Bodard C, Alfsen A, Beaulieu EE. Sex steroid binding plasma protein (SBP). Acta Endocrinol 1970;64: 204.
Anderson DCP. Sex hormone binding globulin. Clin Endocrinol 1974;3:69.
Utian WH. The true clinical features of postmenopause and oophorectomy and their response to oestrogen ther-apy. S Afr Med J 1972; 16:732.
Campbell S. Double-blind psychometric studies oh: üië;y effects of natural estrogens on postmenopausal women. In: Campbell S, ed. The management of the menopause and postmenopausal years. Baltimore: University Park Press, 1976:149.
Thomson J, Oswald I. Effect of oestrogen on the sleep, mood and anxiety of menopausal women. Br Medj 1977; 2:1317.
George GCW, Beaumont PJV, Beardwood CJ. Effects of exogenous estrogens on minor psychiatric symptoms in postmenopausal women. S Afr Medj 1973:47:2387.
Sherwin BB, Gelfand MM. Effects of parenteral admin-istration of estrogen and androgen on plasma hormone levels and hot flushes in the surgical menopause.. AM J . OBSTET GYNECOI. 1984:148:552.
MeLwen BS. Neural gonadal steroid actions. Science 1981;211:1303.